Skip to contentSkip to conversation

generic/556: add a check to make sure ext4 supports encrypted casefolding

PATCH In review Theodore Ts'o 3 participants 4 messages Jun 21, 2021 – Apr 21, 2022

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 generic/556: add a check to make sure ext4 supports encrypted casefolding

Some older kernels support ext4 file systems with encryption enabled, and with casefold enabled, but not file systems have both encryption *and* casefolding enabled. On those kernels, generic/556 will fail. Fortunately, we can test if ext4 supports encrypted casefold via the presence of /sys/fs/ext4/features/encrypted_casefold.

---
 common/casefold   | 17 +++++++++++++++++
 tests/generic/556 |  1 +
 2 files changed, 18 insertions(+)
common/casefold +17 -0
index 9172d818..7b76b986 100644
@@ -43,6 +43,23 @@ _require_scratch_casefold()
 	_require_command "$LSATTR_PROG" lsattr
 }
 
+_require_encrypted_casefold ()
+{
+    case $FSTYP in
+	ext4)
+	    if test ! -f /sys/fs/ext4/features/casefold ; then
+		_notrun "casefolding not supported"
+	    fi
+	    if test ! -f /sys/fs/ext4/features/encryption ; then
+		_notrun "file system encryption not supported"
+	    fi
+	    if test ! -f /sys/fs/ext4/features/encrypted_casefold ; then
+		_notrun "encrypted casefolding not supported"
+	    fi
+	    ;;
+	esac
+}
+
 _scratch_mkfs_casefold()
 {
 	case $FSTYP in
tests/generic/556 +1 -0
index 3145188c..7916a08e 100755
@@ -16,6 +16,7 @@ status=1 # failure is thea default
 . ./common/attr
 
 _supported_fs generic
+_require_encrypted_casefold
 _require_scratch_nocheck
 _require_scratch_casefold
 _require_symlinks
Signature · 1 line
2.31.0
#1 lore raw
Quoted 22 lines · Theodore Ts'o

Some older kernels support ext4 file systems with encryption enabled, and with casefold enabled, but not file systems have both encryption *and* casefolding enabled. On those kernels, generic/556 will fail. Fortunately, we can test if ext4 supports encrypted casefold via the presence of /sys/fs/ext4/features/encrypted_casefold.

---
 common/casefold   | 17 +++++++++++++++++
 tests/generic/556 |  1 +
 2 files changed, 18 insertions(+)
common/casefold +2 -0
index 9172d818..7b76b986 100644
@@ -43,6 +43,23 @@ _require_scratch_casefold()
 	_require_command "$LSATTR_PROG" lsattr
 }
 
+_require_encrypted_casefold ()
+{

common/casefold _require_scratch_casefold() · line 57

Quoted 10 lines
common/casefold +10 -0
+    case $FSTYP in
+	ext4)
+	    if test ! -f /sys/fs/ext4/features/casefold ; then
+		_notrun "casefolding not supported"
+	    fi
+	    if test ! -f /sys/fs/ext4/features/encryption ; then
+		_notrun "file system encryption not supported"
+	    fi
+	    if test ! -f /sys/fs/ext4/features/encrypted_casefold ; then
+		_notrun "encrypted casefolding not supported"

Dumb question: Are there kernels that support encrypted casefold but not those two components separately?

--D

Quoted 23 lines
common/casefold +5 -0
+	    fi
+	    ;;
+	esac
+}
+
 _scratch_mkfs_casefold()
 {
 	case $FSTYP in
tests/generic/556 +1 -0
index 3145188c..7916a08e 100755
@@ -16,6 +16,7 @@ status=1 # failure is thea default
 . ./common/attr
 
 _supported_fs generic
+_require_encrypted_casefold
 _require_scratch_nocheck
 _require_scratch_casefold
 _require_symlinks
Signature · 1 line
2.31.0
#2 lore raw
In reply to Theodore Ts'o

tests/generic/556 status=1 # failure is thea default · line 19

Quoted 9 lines · Theodore Ts'o
tests/generic/556 +1 -0
index 3145188c..7916a08e 100755
@@ -16,6 +16,7 @@ status=1 # failure is thea default
 . ./common/attr
 
 _supported_fs generic
+_require_encrypted_casefold

This isn't an encrypt+casefold test though, but rather just a casefold test. It only becomes an encrypt+casefold test when $MOUNT_OPTIONS includes test_dummy_encryption.

I think we shouldn't update the test itself, but rather make _require_scratch_casefold() call _require_encrypted_casefold() if test_dummy_encryption is in $MOUNT_OPTIONS.

- Eric

#3 lore raw
Quoted 15 lines · Eric Biggers

tests/generic/556 status=1 # failure is thea default · line 19

Quoted 9 lines · Theodore Ts'o
tests/generic/556 +1 -0
index 3145188c..7916a08e 100755
@@ -16,6 +16,7 @@ status=1 # failure is thea default
 . ./common/attr
 
 _supported_fs generic
+_require_encrypted_casefold

This isn't an encrypt+casefold test though, but rather just a casefold test. It only becomes an encrypt+casefold test when $MOUNT_OPTIONS includes test_dummy_encryption.

Quoted 4 lines

I think we shouldn't update the test itself, but rather make _require_scratch_casefold() call _require_encrypted_casefold() if test_dummy_encryption is in $MOUNT_OPTIONS.

Hi Ted, just to follow up on this patch which never got merged:

I tried to reproduce the test failure with the 5.4 and 5.10 kernels, whose ext4 supported casefold but not encrypt+casefold. However, it does *not* reproduce with the "ext4/encrypt" configuration of kvm-xfstests, since _require_scratch_casefold() already detects that the filesystem cannot mounted with both the encrypt and casefold features:

if ! _try_scratch_mount &>>seqres.full; then
_notrun "kernel can't mount filesystem with the encoding set by userspace"
fi

I think the problem is that you were running xfstests with test_dummy_encryption but without adding "-O encrypt" to EXT_MKFS_OPTIONS. That sort of works because of the following code in __ext4_fill_super() in the kernel which force-enables the encrypt feature if the test_dummy_encryption mount option was given:

	if (DUMMY_ENCRYPTION_ENABLED(sbi) && !sb_rdonly(sb) &&
	    !ext4_has_feature_encrypt(sb)) {
		ext4_set_feature_encrypt(sb);
		ext4_commit_super(sb);
	}

However, I don't think ext4 should be doing this. The kernel should generally not be messing with the feature flags, and this code can force-enable 'encrypt' in cases where e2fsprogs would *not* allow it (e.g. the encrypt+casefold situation being encountered here). Also, f2fs doesn't allow this.

So I am planning to send a kernel patch to remove the above code from ext4.

- Eric